Differentiating Phonetically and Phonologically Conditioned Sound Change
Josef Fruehwald, University of Pennsylvania
LSA, January 5, 2013
Outline
Most sound changes exibit conditoning. Some sound changes appear to be conditioned.
- Why is this important to phonologists, phoneticians, and sociolinguists?
- Phonologization of phonetic processes
- Sociolinguistic sensitivity
- What methodologies can we use to make this distinction?
Phonetics and Phonology
Distinguishing the two
Is it possible to distinguish between phonetic effects (universial acoustic and/or gradient coarticulation) and categorical phonological selection? This is, of course, a long standing question. (e.g. Cohn, 1993; inter alia)
Diachrony and Phonology
An increasing number of phonologists point to diachronic sound change and the reanalysis of phonetic coarticulation as the source of phonological generalizations.
- Phonologization (Hyman,1976; Bermúdez-Otero, 2007)
- Maximal Dispersion (Boersma & Hamann, 2008)
- Markedness (Blevins, 2004)
Phonologization
It has been proposed by various researchers that most phonological processes enter grammars as reanalyses of phonetic processes (Hyman, 1976), and it has been suggested that this process of phonologization is the source of phonological naturalness, and markedness (Blevins, 2004).
Phonologization as a gradual process.
- Variation strictly due to production/perception errors.
- Errors accumulate.
- Reanalysis as a phonological process.
(Ohala, 1981)
Sociolinguistic Variation
In Philadelphia, sociolinguistic variation has been reported on for the pre-voiceless allophone of /ay/ (backing it indexes masculinity, and toughness (Conn, 2005; Wagner, 2007)), but not for other allophones of /ay/.
The pre-nasal allophones of /aw/ is markedly more raised and fronted, but has not exhibited sociolinguistic differentation distinct from other /aw/ allophones.
Proposal: This may be because /ay0/ is phonologically differentiated from /ay/, but /awN/ is a phonetic variant.
Gaps to Fill
Change in Progress
To my knowledge, phonologization has not been observed in progress.
Sociolinguistic Utilization
More sociolinguistic work is necessary to see if phonological allophones are utilized differently from phonetic variants.
Data
Philadelphia Neighborhood Corpus
Corpus Develoment
Data used here
- Data from
308 White speakers.
- Interviewed between
1973 and 2010.
- Born between
1888 and 1991.
Case Studies
The effect of nasals on /aw/
|
aw
|
awN
|
total
|
6382
|
5494
|
11876
|
A following nasal favors the direction of the change.
Case Studies
The effect of /l/ on /ow/
|
ow
|
owL
|
total
|
5629
|
1946
|
7575
|
A following /l/ blocks fronting.
Case Studies
The effect of voiceless obstruents on /ay/
|
ay
|
ay0
|
total
|
12131
|
18252
|
30383
|
/ay/ only raises pre-voiceless.
The Trajectories of Change
Background Assumption
- Continuous phonetic shifts in vowels like we examine here are due to shifts in phonetic implementation of surface phonological representations.
Proposal
- If two variants of vowel are phonologically differentiated, they will have two independent targets of phonetic implementation, which may change independently over time.
- If two variants of a vowel are due to a coarticulary, or other phonetic process which happens after the phonology-phonetics interface, their trajectories over time will be yoked, thus parallel.
- Phonologization can be identified by two variants sharing a parallel trajectory at first, followed by divergence.
The Trajectories of Change
Phonological Differentiation
- Phonology
- Phonology-Phonetics Interface
- Vx → Low F2
- Vy → Higher F2
Targets are independent.
The Trajectories of Change
Phonetic Effect
- Phonology
- Phonology-Phonetics Interface
- Coarticulation/Gestural Phasing/Natural Acoustics
Target of VC is yoked to V.
The Trajectories of Change
Consequences
Phonological
The trajectories are independent.
Phonetic
The trajectories are necessarilly parallel.
The Trajectories of Change
Phonological
Divergent trajectories have different rates of change.
Phonetic
Parallel trajectories have the same rate of change.
Trajectories of Change
The easiest way to test for parallelity is to fit straight lines for each variant, and compare the slopes. But, there are important non-linear patterns in these case studies.
Moreover, we’d expect phonologization to exhibit trajectories that are initially parallel, but then diverge.
Modeling the Rate of Change
What I did instead was model the rate of change at ever year, in terms of the size of year-over-year changes.
- Custom written Hierarchical Bayesian model in Stan.
- Estimated via Hamiltonian Monte Carlo.
- Uninformative (flat) priors.
- Allowed for by-speaker and by-word random effects.
- Rate of Change estimated from the data.
- Latent variable δ
- For every year, the amount of change from year Yi to Yi+1.
- Estimated phonetic target for year Yi
- The sum of δ from δ1888 to δi
- Fit using b-splines.
Results
/aw/: Trajectory and Rate of Change Estimates
- [aw] is reliably fronting and raising
- [aw] fronting and raising peaks
- [aw] begins to lower and back, reversing the earlier change.
Results
/aw/: Rate of Change Differences
Conclusion
[aw] and [awN] do not have reliably different rates of change at any point during the 20th century, meaning their phonetic targets appear to be yoked together, even in the reversal of the change.
Results
/ow/: Trajectory and Rate of Change Estimates
Since men barely participated in this change, I’ll be looking exclusively at women.
- [ow] is reliably fronting.
- [ow] fronting peaks
- [ow] begins to back.
Results
/ow/: Rate of Change Differences
Conclusion
- Earliest date of positive [ow] rate of change:
- Earliest date of a reliable difference in [ow] and [owL] rate of change:
[owL] was categorically excluded from undergoing the change from its outset, making this a phonologicallty conditioned sound change.
Results
/ay/: Trajectory and Rate of Change Estimates
Due to weak sex differentiation, this analysis was pooled over men and women.
- [ay0] is reliably raising.
- [ay0] raising peaks.
Results
/ay/: Rate of Change Differences
Conclusion
- Earliest date of positive [ay0] rate of change:
- Earliest date of a reliable difference in [ay] and [ay0] rate of change:
[ay0] was categorically excluded from undergoing the change from its outset, making this a phonologicallty conditioned sound change.
Conclusions
We have seen
- A sound change where phonetic conditioning remained constant across the 20th century.
- Two sound changes which where categorically, phonologically conditioned from their very outset
We have not seen
- Any evidence of gradual phonologization for /ow/ and /ay/.
Conclusions
By bringing together a model of the phonology-phonetics interface, and a theory of phonetic changes, we have produced a novel methodology for distinguishing between phonological and phonetic effects.
Phonological proccesses produce independent targets, and potentially divergent trajectories of change.
Phonetic processes produced yoked targets, thus necessarilly parallel trajectories of change.
References
- Bermúdez-Otero, R. (2007). 21 Diachronic Phonology. In P. de Lacy (Ed.), The Cambridge Handbook of Phonology (pp. 497–517). Cambridge: Cambridge University Press.
- Blevins, J. (2004). Evolutionary phonology: the emergence of sound patterns. Cambridge University Press.
- Boersma, P., & Hamann, S. (2008). The evolution of auditory dispersion in bidirectional constraint grammars. Phonology, 25(2), 217–270.
- Cohn, A. (1993). Nasalisation in English: Phonology or Phonetics. Phonology, 10(1), 43–81.
- Conn, J. (2005). Of `moice’ and men: The evolution of a male-led sound change. University of Pennsylvania.
- Hyman, L. (1976). Phonologization. In A. Juilland, A. M. Devine, & L. D. Stephens (Eds.), Linguistic studies offered to Joseph Greenberg on the Occasion of His Sixtieth Birthday (Vol. 4, pp. 407–418). Anima Libri.
- Kruschke, J. K. (2011). Doing Bayesian Data Analysis: A Tutorial with R and BUGS. Oxford: Elsevier.
- Labov, W. (1994). Principles of linguistic change. Volume 1: Internal Factors. Blackwell.
- Labov, W. (2001). Principles of linguistic change. Volume 2: Social Factors. Language in Society. Oxford: Blackwell.
- Ohala, J. (1981). The listener as a source of language change. In C. S. Masek, R. A. Hendrick, & M. F. Miller (Eds.), Papers from the Parasession on Language and Behavior. Chicago Linguistics Society.
- The Philadelphia Neighborhood Corpus (2011).
- Rosenfelder, I., Fruehwald, J., Evanini, K., & Yuan, J. (2011). FAVE (Forced Alignment and Vowel Extraction) Program Suite.
- Stan Development Team. (2012). Stan: A C++ Library for Probability and Sampling.
- Wagner, S. E. (2007). We act like girls and we don’t act like men: The use of the male-associated variable (ay0) in South Philadelphia. In T. Scheffler, J. Tauberer, A. Eilam, & L. Mayol (Eds.), Penn Working Papers in Linguistics (Vol. 13.1, pp. 393–406). Philadelphia: University of Pennsylvania Press.